Abstract: Numerous researchers and Web development practitioners have proposed various techniques and instruments for measuring usability however, there is still no consensus on how to measure eCommerce website usability. Bearing in mind the call for a better understanding of the usability effects on consumer behavior this study has two research objectives: (1) to propose and validate website usability constructs that mainly captures human perceptions on website usability and (2) to propose and validate a new Lévy-flight Bat Algorithm (LBA) for training neural networks and to provide a comparison study with another relevant algorithm. To provide evidence in support of the first question, this study will create valid and reliable instruments for measuring website usability. To provide evidence in support of the second question, this study will investigate the relationship between the performances of both algorithms. Numerical studies and results suggest that the proposed Lévy-flight Bat algorithm is superior to existing backpropagation algorithms. Finally, we discuss the implication of the results and suggestions for further research.